In [1]:
import pandas as pd
import folium
import geojson
from matplotlib import pyplot as plt
import numpy as np
from PIL import Image
import os
import time
from selenium import webdriver
from colour import Color
Used libraries:
- folium
- geojson
- PIL
- os
- selenium
- colour
To crop images chromedriver.exe must be placed in the executable directory
Load data¶
In [2]:
grouped=pd.read_csv('grouped.csv') #.drop(columns=['trip_duration', 'dropoff_datetime'])
In [3]:
grouped.head()
Out[3]:
Load regions coordinates¶
In [4]:
regions=pd.read_csv('regions.csv', sep=(';'))
regions.head()
Out[4]:
Regions with zero trips¶
In [5]:
# The full number of cells is
regions.shape[0]
# Calculate the number of cells with no one trip
number_of_zero_trips=regions.shape[0]-grouped.shape[1]
print('The number of of cells with no one trip is '+str(number_of_zero_trips))
NYC static map visualization¶
Folium map generator¶
In [6]:
def generateBaseMap(default_location=[ 40.748817, -73.985428], default_zoom_start=11):
base_map = folium.Map(location=default_location, tiles='cartodbpositron',
width=900,
height=880, control_scale=True,
zoom_start=default_zoom_start,
no_touch=True,zoom_control=True )
return base_map
Create geojson file¶
In [7]:
def data2geojson(df):
features = []
insert_features = lambda X: features.append(
geojson.Feature(geometry = geojson.Polygon([[(X['west'], X['south']),
(X['east'], X['south']),
(X['east'], X['north']),
(X['west'], X['north'])]]),
id = str(X.name),
properties=dict(name=X["region"],
description=X["region"])))
df.apply(insert_features, axis = 1)
feature_collection = geojson.FeatureCollection(features)
dumps = geojson.dumps(feature_collection)
return geojson.loads(dumps)
In [8]:
geo_data = data2geojson(regions)
Folium map generation¶
In [9]:
base_map = generateBaseMap()
# set grid to map
folium.Choropleth(geo_data = geo_data,
fill_color = '#00000000',
line_opacity=0.2).add_to(base_map)
# set marker of Empire State Building
folium.Marker([40.748817, -73.985428]).add_to(base_map)
# set text Empire State Building
folium.map.Marker(
[40.748817, -73.985428],
icon=folium.features.DivIcon(
icon_size=(200,36),
icon_anchor=(0,0),
html='<div style="font-size: 9pt; color: Maroon"><b>%s</b></div>' % 'Empire State Building',
)
).add_to(base_map)
# visualize map
base_map
Out[9]:
Save and convert map to png format¶
In [10]:
def save_map(geomap):
# Save map to html
geomap.save('map.html')
# Convert to png image format
delay=5
fn='map.html'
tmpurl='file://{path}/{mapfile}'.format(path=os.getcwd(),mapfile=fn)
base_map.save(fn)
browser = webdriver.Chrome()
browser.get(tmpurl)
time.sleep(delay)
browser.save_screenshot('map.png')
# Crop image and save
im = Image.open('map.png')
width, height = im.size
left = 54
top = 116
right = width-left-6
bottom = height
im = im.crop((left, top, right, bottom))
im.save('map_in_png.png')
browser.quit()
In [11]:
save_map(base_map)
Static map visualization function¶
In [12]:
# map visualization plot
def map_visualization(longitude, latitude,s=7):
x_min=-74.25559
x_max=-73.70001
y_min=40.49612
y_max=40.91553
img = plt.imread('map_in_png.png')
fig, ax = plt.subplots(figsize=(12, 12))
fig.set_facecolor('.96')
ax.imshow(img, extent = [x_min, x_max, y_min,y_max])
ax.scatter(x=longitude, y=latitude, c='g' , s=s, label='Taxi pickup point')
ax.legend(loc='upper left', fontsize=16)
plt.title('New York city map', size=20, y=1.03)
plt.xlabel('Longitude')
plt.ylabel('Latitude')
ax.xaxis.grid(True, which='major', linewidth=0.1,
color='black', linestyle='--')
ax.yaxis.grid(True, which='major', linewidth=0.1,
color='black', linestyle='--')
Static map¶
In [13]:
map_visualization(-73.985428, 40.748817)
Check regions on static map¶
In [14]:
long_mean=(regions.west+regions.east)/2
lat_mean=(regions.south+regions.north)/2
In [15]:
map_visualization(long_mean, lat_mean)
Count trips heatmap in may of 2016 from each region¶
Trips count from each region¶
In [16]:
grouped_count=grouped.sum().to_frame().reset_index().rename(columns={'index': 'region', 0:'count'})
grouped_count['count']=grouped_count['count'].astype('int')
grouped_count
Out[16]:
Heat map visualization¶
In [17]:
# create folium heatmap
base_map = generateBaseMap()
choropleth=folium.Choropleth(
geo_data = geo_data,
data=grouped_count,
columns=['region', 'count'],
key_on='feature.id',
fill_color='YlOrBr',
nan_fill_color='#00000000',
fill_opacity=0.6,
line_opacity=0.2,
smooth_factor=3,
legend_name='Taxi trips count',
reset=True).add_to(base_map)
# set marker of Empire State Building
#folium.Marker([40.748817, -73.985428]).add_to(base_map)
# add popup of regions to determine the number of region
style_function = "font-size: 15px; font-weight: bold"
choropleth.geojson.add_child(
folium.features.GeoJsonTooltip(['name'], style=style_function, labels=False))
# set text Empire State Building
folium.map.Marker(
[40.748817, -73.985428],
icon=folium.features.DivIcon(
icon_size=(200,36),
icon_anchor=(0,0),
html='<div style="font-size: 9pt; color: Maroon"><b>%s</b></div>' % 'Empire State Building',
)
).add_to(base_map)
folium.LayerControl().add_to(base_map)
Out[17]:
In [18]:
base_map
Out[18]:
In [20]:
save_map(base_map)
In [21]:
trips_data = np.array(grouped_count['count'])
coord_data = np.array(regions)
In [22]:
# create static heatmap
# define colors
color_1 = 'LightYellow'
color_2 = 'Brown'
color_1 = Color(color_1)
grad = list(color_1.range_to(Color(color_2), int(trips_data.max())))
# Heatmap
fig, ax = plt.subplots(figsize=(12, 1.5))
fig.set_facecolor('.96')
trips_values = list(range(0, int(trips_data.max())+1, 500))
colorbar = np.array([[tuple([int(round(j*255, 0)) for j in grad[int(i)-1].rgb]) for i in trips_values] for _ in range(100)])
ax.imshow(colorbar)
plt.yticks([])
ticks = list(range(len(trips_values)))
plt.xticks(ticks=ticks[::100] + [ticks[-1]], labels=trips_values[::100] + [trips_values[-1]])
plt.title('Taxi trip counts heatmap', y=1.01, fontsize=20)
plt.show()
map_visualization(-73.985428, 40.748817, 70)
Filter and count average number of trips from each region¶
In [23]:
grouped_mean=grouped.mean().to_frame().reset_index().rename(columns={'index': 'region', 0:'mean'})
grouped_mean['mean']=grouped_mean['mean'].astype('int')
grouped_mean=grouped_mean[grouped_mean['mean']>=5]
grouped_mean
Out[23]:
Heatmap average number of trips from each region¶
Heatmap visualization¶
In [24]:
# create folium heatmap
base_map = generateBaseMap()
choropleth=folium.Choropleth(
geo_data = geo_data,
data=grouped_mean,
columns=['region', 'mean'],
key_on='feature.id',
fill_color='YlOrBr',
nan_fill_color='#00000000',
fill_opacity=0.6,
line_opacity=0.2,
smooth_factor=3,
legend_name='Taxi trips count',
reset=True).add_to(base_map)
#set marker of The Statue of Liberty
folium.Marker([40.689247, -74.044502]).add_to(base_map)
# add popup of regions to determine the number of region
style_function = "font-size: 15px; font-weight: bold"
choropleth.geojson.add_child(
folium.features.GeoJsonTooltip(['name'], style=style_function, labels=False))
folium.TileLayer('openstreetmap').add_to(base_map)
# set text Empire State Building
folium.map.Marker(
[40.689247, -74.044502],
icon=folium.features.DivIcon(
icon_size=(200,36),
icon_anchor=(0,0),
html='<div style="font-size: 9pt; color: Maroon"><b>%s</b></div>' % 'The Statue of Liberty',
)
).add_to(base_map)
folium.LayerControl().add_to(base_map)
Out[24]:
In [25]:
base_map
Out[25]:
The number regions with more than 5 average trips per hour¶
In [26]:
print('The number of regions after filtering is '+str(len(grouped_mean)))
In [28]:
grouped_mean.region.unique()
Out[28]:
In [29]:
len(grouped_mean.region)
Out[29]:
In [ ]: